Agentic AI Expert @ IBM
Designing multi-agent systems for enterprise clients on watsonx + LangGraph.
- ✓Multi-agent orchestration for enterprise workflows
- ✓RAG pipelines grounded in client data, with citations
Udit Narayana — full-stack engineer and agentic-AI specialist. I’ve sold it, operated it, and now I build it, end to end.
Sales taught me to listen. Operations taught me to deliver. Engineering lets me do both — at scale.
Today I build agentic AI: systems that plan, retrieve, remember and act — and stay honest when they don’t know.
Pick a category. Bars show depth — the longer the bar, the more production miles.
It’s exactly why my agents solve business problems, not just benchmarks.
Designing multi-agent systems for enterprise clients on watsonx + LangGraph.
Ran warehouse-as-a-service operations at India’s largest e-commerce player.
Consultative edtech sales — where I learned to truly listen to users.
Projects I’m proudest of. Use the arrows to browse, or jump to the full archive at the end.
bcjksbjzsbcajbcsjzkhdbvkzhbkdzlabl l i ggusgus
A retrieval-grounded assistant that answers questions about my work in real time — built on a vector-search backend, streamed token-by-token, and tuned to never hallucinate a project that doesn't exist.
Crucible AI began as an experiment to transform scattered enterprise knowledge into intelligent decision-making. By combining LLMs, secure data pipelines, and automation, it evolved into a platform that turns information into action. 🚀
Browse everything I’ve built, with filters and the story behind each one.
See all projectsMachine learning, distributed systems and human-computer interaction. Capstone: a neural retrieval engine for lecture archives.
Sequence models, attention and transformers — the theory underneath the agents I build today.
Practical notes on agents, retrieval and shipping.
Want more? Every concept I learn goes on lighthouseinfo.com, with search, topics and a newsletter.
More blogs on lighthouseinfo.comHave an idea? I take on a few projects at a time: full-stack builds, AI agents, or rescuing a product that’s stuck. I’ll reply within a day or two.